finished Shoji lamp -pics

[blasterman]

After a few gallons of Epoxy and a lot of cussing with a hand miter saw I finally finished the darn thing. The ambient light this thing puts out is really fantastic. What I need to do now is figure out how to simplify the framing structure so I can more easily build and sell them. I also need to give a technical credit to both my cats who helped supervise the entire project.

 

[McGizmo]

Really nice!!!! :thumbsup:

 

[maxspeeds]

Wow that is beautiful! Can we see pictures of inside the lamp?

 

[blasterman]

When I have a chance I'll take a pic, but there's not much to the inside. It's just 5 bare Cree stars arranged in a wide pattern on a scrap piece of 7" square steel plate. Three warm-whites, two R2 cool whites. I'm using a 350mA dedicated driver.

 

[jeffosborne]

Bravo blasterman! Looking very good. Beautiful soft lighting, and I like the neat patterns of light on the wall and ceiling. Tell us, what is the diffusion material on the sides? And what sort of price are you considering for the sale of this lamp? Cheers, Jeff

 

[ponygt65]

WOW>...that looks great. I really like the design of the wall wash affect. Nice job you guys.

 

[HoopleHead]

Wow, very cool and unique take, nice work!

 

[jason 77]

"After a few gallons of Epoxy and a lot of cussing with a hand miter saw" "What I need to do now is figure out how to simplify the framing structure so I can more easily build and sell them."

Well you could start out with getting a power miter saw "compound miter saw" or like a table saw, since using a hand miter saw is quite tedious! Also make a jig to make the different parts so you can mass produce the peices needed.

I would really like to see the inside of this lamp!

 

[blasterman]

Here's a pic of the inside of the lamp showing how simple it's is. The open bottom even shows my wood floor. Second pic shows the nice diffraction pattern the fabric creates. At some point I want to build a bigger / taller version, but I'll likely use alabaster textured stained glass. The fabric gives a really nice 'look', but it's reason I have to make so many extra wood cuts to hide the edges of the cut fabric. I haven't hemmed' fabric since home-ec in high-school, and would like to keep it that way

 

[jason 77]

The fabric gives a really nice 'look', but it's reason I have to make so many extra wood cuts to hide the edges of the cut fabric. I haven't hemmed' fabric since home-ec in high-school, and would like to keep it that way

You could try using this stuff instead of that fabric, I have some and it cuts like paper, no frayed edges....

http://www.tapplastics.com/shop/product.php?pid=241&

 

[blasterman]

Hmmmm.......that might do the trick!

 

[purduephotog]

Hmmmm.......that might do the trick!

Blasterman-

I remember you saying you had some problems at one point with LEDs dying due to inrush conditions-

I just saw this article on using an inrush resistor thermister to help prevent that-

http://www.ledengin.com/products/appnotes/opguid-LZprod_revA.pdf

Might be worth reading- don't know how it would work in your case, but the idea is sound.

(I'm trying to figure out how to build an AC line driver for 6-8 LedEngin 10W LEDs and came across it on google)

 

[blasterman]

Jeeze, where have you been

I've been killing Crees and generics left and right with a project requiring a couple 700-watt Xitaniums, and was about ready to go back to ol' reliable computer PSU's and ghetto style 10-watt rad-shack resistors. Not very efficient, but I've never lost an LED with a standard 12-volt powersupply or laptop brick this way, but my grave-yard is now at several dozen caused by current regulated supplies.

These things are frikken evil, and as I've learned the 700mA ones are really, really bad. The article you posted confirms this. Lately I've been researching high efficiency, *STABLE* precision voltage sources for LED driving -vs- current regulators because I'm sick of the problem. Cree is the most sensitive to failure with BestHong Pro Lights being the hardiest on high current LED supplies...go figure.

I've also got a lot of E-mail from people complaining about piles of dead
LEDs (and bucks), so I'm not the only one skeptical of the design. A 700mA Xitanium driver when it shorts out without load throws sparks like I'm jumping my car on a Saturday night...yeah...no problem there (not).

...And now you solved it with a $2.50 part Much obliged.

Oh yeah...I already retired the Shoji design. I replaced the box from clear textured glass from a stained glass store, and the improvement was dramatic. LEDs look like firework embers frozen in a glass box, and it's 3x as bright, and it's far easier to build.

 

[wkumtrider]

What happened to the pics?

 

[purduephotog]

Jeeze, where have you been

I've been killing Crees and generics left and right with a project requiring a couple 700-watt Xitaniums, and was about ready to go back to ol' reliable computer PSU's and ghetto style 10-watt rad-shack resistors. Not very efficient, but I've never lost an LED with a standard 12-volt powersupply or laptop brick this way, but my grave-yard is now at several dozen caused by current regulated supplies.

These things are frikken evil, and as I've learned the 700mA ones are really, really bad. The article you posted confirms this. Lately I've been researching high efficiency, *STABLE* precision voltage sources for LED driving -vs- current regulators because I'm sick of the problem. Cree is the most sensitive to failure with BestHong Pro Lights being the hardiest on high current LED supplies...go figure.

I've also got a lot of E-mail from people complaining about piles of dead
LEDs (and bucks), so I'm not the only one skeptical of the design. A 700mA Xitanium driver when it shorts out without load throws sparks like I'm jumping my car on a Saturday night...yeah...no problem there (not).

...And now you solved it with a $2.50 part Much obliged.

Oh yeah...I already retired the Shoji design. I replaced the box from clear textured glass from a stained glass store, and the improvement was dramatic. LEDs look like firework embers frozen in a glass box, and it's 3x as bright, and it's far easier to build.

My pleasure.

It was always in the back of my head as a nagging issue- I just couldn't get the brain to wrap itself around what the solution was. I finally accidentally found the answer...

I suppose putting it somewhere around either the 'driver' page or somewhere else to help others would be good.

Yeah I bet the stained glass lit that up much better- nothing beats transmission.

 

[blasterman]

Link fixed (so much for free web hosting).

Right now I'm seriously ticked at the LED driver makers. This has been a learning curve for me, but it seems logical that their drivers (Xitanium, etc) should incorporate some type of buffer to level off current burst. After all, I'm trying to building lamps and not a stun gun or taser.

They seem to all think their drivers are going to be used with dozens of 150mA LEDs in series and nobody actually uses them with 3watt emitters.

Yeah...the glass version is really a step up. Rippled / textured glass does cool things with bright LEDs behind it.

 

[lonesouth]

You can't tout how wonderful your new lamp is without at least giving us something to oogle...that's just wrong

 

[JohnR66]

I like the pattern thrown on the ceiling using the different color tints of the LEDs. You could make a top piece with interesting shapes to throw unique patterns on the ceiling. This would make an interesting conversation piece. You should try to sell them in some of the art galleries around.

 

[Dave_H]

Right now I'm seriously ticked at the LED driver makers. This has been a learning curve for me, but it seems logical that their drivers (Xitanium, etc) should incorporate some type of buffer to level off current burst. After all, I'm trying to building lamps and not a stun gun or taser.

They seem to all think their drivers are going to be used with dozens of 150mA LEDs in series and nobody actually uses them with 3watt emitters.

Losing expensive LEDs to what appears to be bad drivers is a
bad thing. I don't know all the specifics of current-mode convertors,
mostly deal with voltage-mode where the voltage overshoot is
well-controlled. This is why the PC power supply and resistors
is a safe (if inefficient) method. From the sound of it, the current-
mode driver may be overshooting on turn-on. Can you confirm;
is this a case of the LEDs lighting brighter for a short time just
after power-on or hookup?

I understand current-mode drivers will adjust their output voltage
to the level that the load draws the set current. This takes a finite
amount of time, and if starts at the high end and adjusts down,
sounds like a bad design.

Could you point me to specs of these drivers?

A thermistor in series might work, on the other hand anything
interfering with the driver's ability to regulate current might drive
it in the wrong direction, which might be bad.

Still thinking about it.

Dave

 

[blasterman]

Can you confirm;
is this a case of the LEDs lighting brighter for a short time just
after power-on or hookup?

No....nothing noted there. As I understand it the burst is in milliseconds. When the LEDs burn out they typically don't even light at all.

The best indication of how much power is actually transient is simply shorting out a current regulated supply with no load and noting the crackle and rather long sparks. I've had the same effect and problems with both Xitanium and LightTech drivers, and I would assume their design is very similar. Other people e-mailing me telling me they've had the same issues.

The problem can be averted if the driver is powered up with sufficient load on it, which is the assumed design use. Or, using a decent power resistor in series with the powered driver when used to test off line LEDs. However, for the first time I was able to induce a cascade failure with mixed 3-watt LEDs by putting a couple generic LEDs in series with Crees and K2s. The loss of the flaky LED was obviously enough to fool the driver into pulsing enough current to trash the better LEDs. That has me concerned - and not because I insist a Xitanium be able to normally handle weird loads. The driver is obviously designed for a reasonable load.

My concern here is that if some LEDs are far more sensitive to current burst from this type of driver then there might be a long term delta with failure rates. So far, my most durable installations have been from fixed 12-volt supplies with resistors added. Again, not very efficient, but seems very stable with any LED load I throw at it. The question I have is why not simply alter the second stage trandformer in this type of supply to get the voltage you actually want, and skip everything else. Why have a need then for current regulation?

When I have a chance I'll test the GE Thermistor on a string with LEDs I know have problems with a 700mA supply and see what happens (cough* Satistronics Cough*). Maybe it's a good idea just to put a thermistor inline with a Xitanium supply just to be safe anyways given they are pretty cheap and don't seem to have much runtime resistance to matter much.

This topic might be better served moved to it's own thread.